摘要
实现指针式仪表示数的视觉识别,是监控现场自动化和无人化的重要工作。本文以此为背景,开展了指针式仪表图像预处理工作的研究。首先,设计了四个环节的预处理框架;其次,使用线性模型完成灰度化处理,使用大津法完成阈值分割处理,构建八向连通链码确定表盘区域,采用腐蚀算法去除表盘噪声。实验结果表明,本文设计的预处理方案和相关方法,较好地提取了表盘目标区域,清晰地保留了各种特征,为后续的特征提取工作奠定了基础。
Realization of visual identification for pointer-type instrument is an important work of automation of un-manned monitor. Based on this, preprocessing work is carried out for pointer type meter in this paper. First, a new framework is built for preprocessing including four aspects. Second, the linear model is used in completing gray pro-cessing and Otsu method is used in completing threshold segmentation. Then, eight-connection chain code is built to determine the dial region, and corrosion algorithm is selected to remove noise. Experimental results show that dial re-gion can be extracted and various features can be retained by using the methods proposed in this paper. Our work also built a foundation for the subsequent feature extraction.
出处
《电测与仪表》
北大核心
2013年第S1期129-131,136,共4页
Electrical Measurement & Instrumentation
关键词
指针式仪表
视觉识别
阈值分割
八向连通链码
pointer-type meter,visual recognition,threshold segmentation,eight-connection chain code
